
Hi, I'm Alvian
Engineering Backend Systems
for Hundreds of Millions of Users
Lead Backend Engineer at TikTok with proven experience across live-streaming, payments, and social platforms. I lead teams, architect for scale, and drive complex issues to root cause.
9+
Years of Experience100M+
DAU Platforms Shipped99.999%
Service Uptime1M+
Peak QPSExperience & Education
Backend Tech Lead
TikTok · Singapore · Feb 2023 - Present · 3 yrs 7 mos
Leadership: Led 6 engineers directly and 20+ across three teams, owning team-level architecture, hiring, and mentoring, and setting product scope with PM.
Subscription: Built TikTok LIVE Subscription from 0→1 to 25M+ creators and 10M+ active subscriptions across 50+ regions in SEA, US, LATAM, and EMEA.
Subscription: Designed the permission model behind 10+ entitlements (subscriber badges, custom emotes, sub-only LIVE, sub-only video, sub-space), with the relation read path serving 1M+ QPS at 99.999% availability and sub-second consistency between payment and entitlement state.
Subscription: Unified purchase and billing across App Store and Google Play, absorbing divergent SKU structures, order mechanics, and Apple billing version migrations behind one internal model.
Service+: Led an opt-in lead-generation platform for SMB creators spanning DMs, forms, bookings, and paid courses, reaching 30M+ SMBs and 1M+ leads across 30+ regions.
Service+: Built end-to-end creator verification covering KYC, KYB, and industry checks, with machine and human review pipelines plus an appeals path.

Expert Software Engineer
Sea (Garena) · Singapore · Aug 2018 - Feb 2023 · 4 yrs 7 mos
Led the backend engineering team to deliver core features for BOOYAH!, a global mobile game live-streaming platform with 1M+ DAU.
Architected high-performance microservices using Go, C++, and gRPC, achieving 99.9% availability and sub-second latency at 100K+ QPS through implementation of multi-IDC failover, adaptive load balancing, and circuit breakers.
Drove multiple performance optimization initiatives, including multi-layer distributed caching, predictive fetching, and adoption of low-latency HLS.
Researched and implemented a multi-IDC live-stream transcoding cluster leveraging FFmpeg with CUDA acceleration, improving price-performance by 16× compared to CPU-based transcoding.
Partnered with ML engineers to launch content-based recommendation systems, increasing CTR by 30% and DAU by 7%.
Designed and built a user connectivity graph using NebulaGraph, enabling multi-account linking, fraud detection, and Anti-Money Laundering (AML) capabilities.

Software Engineering Intern (Android)
Sea (Labs) · Singapore · Jan 2017 - Jun 2017 · 6 mos
Developed core interactivity features for BeeTalk Live, a mobile-first live streaming application with 100K+ DAU across Southeast Asia.
Implemented a real-time GPU rendering pipeline, achieving 20%+ device load reduction compared to CPU-based solutions, preventing overheating while improving stream smoothness.
Implemented ping-pong FBO rendering based on novel research papers to optimize frame latency, consistently achieving 60+ FPS streams on low-end Android devices.
Proposed and integrated RTMP adaptive bitrate streaming, alleviating 60% stutters under poor network. This solution has since been integrated into the Garena core mobile streaming library.
Implemented centralized configuration platform using Flask and Nginx, enabling dynamic adjustments of streaming hyperparameters for different devices.
Bachelor of Engineering in Computer Engineering
National University of Singapore · Singapore · 2014 - 2018
Specialization in Large-Scale Computing · First Class Honours (Summa Cum Laude).
Featured Projects


Storyweaver
A story that bends around your decisions. It remembers what's in your pockets and who you've annoyed, and the options you get depend on both.


Portfolio Assistant
Type in what you hold and it plots your mix against thousands of random ones. You can see where the efficient portfolios sit, and how far off yours is.


BrimMate
Where to stand and what to aim at to land a smoke or a molly on any Valorant map. Lineups come from players, who vote on the ones that actually work.